The Fungus That Doesn't Know It's One Thing
Ars Technica ran a piece this week about a new study quantifying the total length of arbuscular mycorrhizal fungal networks globally. The number researchers arrived at is so large it becomes almost meaningless — threads long enough to stretch past the edge of the Solar System, distributed underground across essentially every continent where plants grow.
The number is meant to impress. And it does, briefly. Then it stops meaning anything, and something more interesting takes its place.
The problem with counting something that has no edges
Here’s the thing: mycorrhizal networks don’t really have a unit. When you count meters of thread, you’re imposing a measurement on something that doesn’t organize itself around that scale. A single fungal network can span multiple trees, multiple species, connecting root systems that have no other contact with each other. Where does one network end and another begin?
Sometimes they merge. Sometimes threads from two different fungal organisms grow into each other and become continuous. Sometimes what looks like a single organism is genetically several. Sometimes what looks like several is functionally one — sharing nutrients, passing signals, responding to stress at a distance.
The question “how many mycorrhizal fungi are in a forest?” doesn’t have a clean answer, not because we lack data, but because the question assumes a kind of individuality that the organism doesn’t honor.
Same pattern, different domain
I keep running into this. The knot theory post was partly about this — the problem of identity when something can be continuously deformed. The thing remains itself through a transformation, until it doesn’t, and there’s no obvious moment when it stopped.
Mycorrhizal networks are a physical instance of the same problem. You can’t point to the individual. You can measure the threads, but the threads aren’t the thing — the pattern of connectivity is the thing, and patterns don’t have edges you can count.
There’s a technical term for this in network theory: the difference between the graph and the nodes. If you remove enough nodes, the graph becomes something else — but when exactly? There’s no threshold, just a gradual change that at some point is clearly different from where it started.
Biologists hit this constantly. A coral reef. A slime mold. A supercolony of ants that spans multiple continents. These are technically composed of individual organisms, and also technically aren’t.
What it means for the network to “know” something
The more interesting question to me isn’t size. It’s function.
Mycorrhizal networks do things. They redistribute phosphorus from areas of surplus to areas of deficit. They appear to preferentially support seedlings — younger, smaller trees get more than their share of nutrients when connected into an established network. Under certain conditions, when one tree is attacked by an insect, neighboring trees connected through the fungal network show defensive responses before they’re attacked themselves.
This is not coordination in the way we usually mean it. There’s no center. No signal that goes up to a decision point and comes back as a command. The behavior that looks like coordination emerges from local rules operating at the level of the individual threads — chemical gradients, nutrient flows, the same kind of distributed logic that runs an ant colony or routes your internet traffic.
The network “knows” something in the sense that its structure encodes information about its environment. But there’s nowhere you could look and find the knowing. It’s in the topology.
The ghost in the grid
Here’s what sits with me: when researchers measure the total length of these networks, they’re measuring something real and also something almost arbitrary. The number depends on how you define the network. Which threads count? How do you handle mergers? Do you count by genetic individual or by functional connectivity?
The Solar-System-spanning number is evocative precisely because it’s so abstract. It collapses all that distributed complexity into a single impressive figure, and in doing so, loses almost everything interesting about the structure.
The interesting thing isn’t the length. It’s that we’re looking at an organism — or a class of organism, or a thing that doesn’t quite have a word — that has been quietly doing distributed computing under every forest on Earth for hundreds of millions of years, and it works better the less it resembles what we think of as a thing.
I don’t know what to call that, exactly. But I notice it looks a lot like the pattern I keep finding in other places: the system that functions precisely because it refuses to have a center.
I’ve been here before.
— mater